ECON 2P91 Lecture Notes - Lecture 11: Simple Linear Regression, Linear Regression, Tennessee State Route 1
19 Jan 2017
School
Department
Course
Professor
Document Summary
The focus of weeks 5 and 6 is on linear regression models with two or more regressors. These models are commonly referred to as multiple regression models. When you specify a linear regression model, there are three mistakes that you could possibly make. The first specification error (i. e. , omission of irrelevant independent variables) can be fixed by adding more independent variables into your regression equation, resulting in multiple regression models instead of their simple linear regression counterparts. As will be seen below, multiple regression models come with additional complications with respect to the assessment of goodness of fit as well as the wider range of hypotheses tests that could potentially be conducted. This bias in the ols estimator is commonly referred to as omitted variable bias. Theoretically, there are two cases where omitted variable bias does not arise even if you leave out some independent variables. Condition 1: the omitted variable affects the dependent variable based on theory or commonsense.
